Rsync over ssh, or something else?

"You need an rsync command on the server, but you don't need to run a daemon.

Get an rsync binary that works on the server, put it somewhere in your home, and add this flag to your command line: --rsync-path=/home/user/path/to/rsync .

If you don't want to copy rsync to the servers, you can use scp, or sshfs.

sshfs user@host ~/sync/user-host
rsync -av ~/local-dir ~/sync/user-host/remote-dir

In that case rsync will run completely locally, but the changes will be propagated to the server."

"There is not much point using rsync without running an rsync daemon on the remote host, in spite of the creative sshfs solution proposed above. The whole idea of efficiency in rsync is to have the two instances compare their local copies of the dataset by calcuating checksums locally, comparing them over the network, and only transmitting the actual data for those parts of those files which actually differ.

If you use sshfs or any network mount, it's only hiding the fact that it has to copy the entire dataset across the network from the remote host in order to calculate the checksums locally, and then transmit the differences on top of that. It would be quicker and cheaper to delete everything locally and 'scp -r' the lot."